ISSUE 358  ·   October 19, 2021

Trends

China - Surveillance state or way of the future?

In China, the state collects massive amounts of data from its willing citizens. Critics call it "the most ambitious Orwellian project in human history" but there are many practical benefits and people who play by the rules are rewarded. Check it out. Is this the future you want?
YouTube | DW Documentary (45 minutes)

Sponsored Link

Webinar: Build customer-oriented applications using third-party data

In this virtual session, AWS Data Exchange will host a discussion with thought leaders from companies such as Foursquare and NextDoor. They will share real-world examples on leveraging datasets to create a customer-centric strategy and improve business outcomes.

Tutorials, Projects & Opinions

Statistical problems found when studying Long Covid

Statistical tests need to be paired with proper data and study design to yield valid results. A recent review paper on Long Covid in children provides a useful example of how researchers can get this wrong. This post uses causal diagrams to decompose the problem and illustrate where errors were made.
fast.ai | Jeremy Howard

How wavelets allow researchers to understand data

This introduction to wavelets shows how researchers use them to discover patterns in scientific data by analyzing entire data sets at once.
Quanta Magazine | Alexander Hellemans

Machine learning is not nonparametric statistics.

Statisticians often claim that machine learning is nothing more than nonparametric statistics. But in reality, statistics and machine learning have different goals and are used to pursue different questions.
Ben Recht

Breathing K-Means 

Breathing K-Means is an approximation algorithm for the k-means problem that generally outperforms scikit-learn's k-means++. This collection of Jupyter Notebooks shows how it works. For details, see the Breathing K-Means paper.
GitHub | Bernd Fritzke

Neural Networks from Scratch

Great introduction to neural networks! This is complete with an interactive playground that shows how each layer works as you go. It's easy to follow and is very well done.
Allison George

Webinar: How to deliver actionable insights at scale

Oct 28 |  2PM ET / 11AM PT
Learn how to deliver analysis-ready information for actionable insights across your organization with speakers from Allstate and Jackson National Life. 
// sponsored

Career

What to Learn

Everyone has unique strengths and weaknesses but if you study successful people in different fields, you'll see a common approach to how they choose which skills to develop. Learning takes time and there are opportunity costs to learning the wrong things. So how should you choose what to learn?
Dan Luu

Data Visualization

Generative art resources in R

Great post for getting started with making generative art using R. Covers a wide variety of art forms, such as watercolor simulation, trees, particles, cellular automata, flow fields and more. Follow the links!
Danielle Navarro 

D3 Charts - A new approach to D3’s example charts

D3 Charts is a new generation of D3 examples that are structured as functions of data that accept options and are easily reused elsewhere. The collection includes standard D3 chart components, such as bar charts, beeswarm plots, candlestick charts, choropleths, and lots more.
Observable

How did you like this issue of Data Elixir?

👎    1     2     3     4     5   👍

Was this email forwarded to you?  Sign up here >>